Prosocial Benefits of Feeling Free: Disbelief

in Free Will Increases Aggression and

Reduces Helpfulness
Roy F. Baumeister
E. J. Masicampo
Florida State University
C. Nathan DeWall
University of Kentucky

Laypersons belief in free will may foster a sense of

thoughtful reflection and willingness to exert energy,
thereby promoting helpfulness and reducing aggression, and so disbelief in free will may make behavior
more reliant on selfish, automatic impulses and therefore less socially desirable. Three studies tested the
hypothesis that disbelief in free will would be linked
with decreased helping and increased aggression. In
Experiment 1, induced disbelief in free will reduced
willingness to help others. Experiment 2 showed that
chronic disbelief in free will was associated with
reduced helping behavior. In Experiment 3, participants induced disbelief in free will caused participants
to act more aggressively than others. Although the
findings do not speak to the existence of free will, the
current results suggest that disbelief in free will
reduces helping and increases aggression.
Keywords:

free will; agency; aggression; helping; prosocial

behavior

elief in free will seems widespread and intuitive.

Almost every person every day has the subjective
impression of making a choice in which more than one
outcome is possible. The most influential religious beliefs
in Western culture give prominent emphasis to doctrines
of free will, assuming that human individuals can freely
choose whether to perform virtuous or sinful actions and
even stating that eternal judgment of individual souls
rests on the choices they make. Likewise, the legal system
allocates guilt and punishment differentially based on
whether the rule breaker could have acted differently,

such that perceived reductions in the capacity for free

choice (including external pressures, lack of awareness,
mental illness, or intense emotion) constitute valid reasons for reduced punishment or even acquittal.
Intellectuals and scientists, however, seem rather less
uniformly comfortable with the idea of free will than the
general public. Many scientists regard the belief in free will
as untenable if not downright absurd. Some scientific writers such as Crick (1994) insist that all human thought and
action are strictly determined by brain processes. Although
not explicitly siding with them, Wegner (2002) summarized the opposition to free will as embodying the assumption that only bad scientists could believe such a thing.
To be sure, the impossibility of free will cannot be
proven either empirically or conceptually. Scientists
mainly oppose it because they think that determinism is
a necessary assumption rather than a proven fact. The
human mind understands and explains events in terms
of deterministic causality (e.g., Kant, 1787/1967), but
perhaps that says more about the way the mind operates
than about reality per se. Among disciplines, philosophy
lays claim to the most rigorous thinking, and many

Authors Note: We are grateful for a grant from the Templeton

Foundation. Please address correspondence to Roy Baumeister,
Department of Psychology, Florida State University, Tallahassee, FL
32306; e-mail: baumeister@psy.fsu.edu, or C. Nathan DeWall,
Department of Psychology, University of Kentucky, Lexington, KY
40506-0044; e-mail: nathan.dewall@uky.edu.
PSPB, Vol. 35 No. 2, February 2009 260-268
DOI: 10.1177/0146167208327217
2009 by the Society for Personality and Social Psychology, Inc.

260
Downloaded from psp.sagepub.com at University of Tokyo on September 3, 2011

Baumeister et al. / PROSOCIAL BENEFITS OF FEELING FREE

philosophers have retreated to a so-called compatibilism that asserts that free will, or at least freedom of
action, can be reconciled with deterministic causation
(see Kane, 2002).
The present investigation does not take a position on
the reality of free will, nor is it even directly concerned
with whether free will exists. Rather, it sought to investigate the consequences of belief in free will. If freedom of
action is deeply embedded in Western culture and its
social systems, what might be the effects of seeking to
undermine and discredit that belief? If one assumes that
cultural beliefs are not arbitrary but rather are functional
in most cases, the belief in free will may be expected to
benefit society in some ways; therefore, undermining that
belief might produce antisocial tendencies.
The point of departure for this investigation was a
recent investigation by Vohs and Schooler (2008). In different studies, they induced disbelief in free will in several
ways, such as by having participants read Cricks (1994)
vehement lampooning of the idea of free will or by having them follow a Velten-like procedure in which they
read statements asserting the lack of human freedom of
action. Afterward, these participants were more willing
than control participants to cheat on a test (and, in one
case, effectively steal money from the researchers). Those
authors speculated that to undermine peoples belief in
free will could have socially costly consequences in terms
of increasing antisocial behavior.
The present experiments sought to extend the Vohs
and Schooler (2008) findings into a broader context,
namely, helping and aggression. If helping decreases and
aggression increases, that would emphatically support
and extend Vohs and Schoolers conclusion that disbelief in free will reduces socially desirable behavior.
Cheating, after all, can be regarded as a strategy for
improving (ostensible) performance, and so one could
argue that what drove the increase in cheating was a
desire to achieve as high a score as possible. In contrast,
if disbelief in free will were shown to increase aggression and reduce helpfulness, then achievement-based
explanations would largely be ruled out.
Why would disbelief in free will increase aggression
and reduce helping? A belief in free will may be crucial
for motivating people to control their automatic impulses
in favor of more prosocial forms of behavior. The automatic response for most people seems to be to act selfishly and without concern for strangers, so that a
significant amount of self-control and mental energy is
required to override this default to help others (DeWall,
Baumeister, Gailliot, & Maner, in press) and to restrain
aggressive impulses (DeWall, Baumeister, Stillman, &
Gailliot, 2007). A disbelief in free will may therefore
undermine this prosocial form of control. Consistent

261

with that line of reasoning, Vohs and Schooler (2008)

proposed that disbelief in free will serves as a subtle cue
that exerting volition is futile and thereby gives people
permission not to bother. The idea of not bothering to
exert volition appeals to people insofar as volition in the
form of self-control and choice requires exertion and
depletes energy (Baumeister, Bratslavsky, Muraven, &
Tice, 1998; Gailliot et al., 2007; Vohs et al., 2008). In a
sense, then, making people disbelieve in free will may
serve as a nonconscious prime to act in relatively automatic ways, which would thus include enacting impulses
rather than exerting control and restraint.
We examined two related possibilities, as well. One
was that inducing belief in free will stimulates a conscious
feeling of being active and energetic, thereby making
people feel like they want to exert control. Another is that
belief in free will supports (and disbelief undermines) a
sense of personal responsibility and accountability.
Feelings of responsibility and accountability may make
people feel that they ought to behave in socially desirable
ways, such as performing prosocial acts of helping and
restraining antisocial impulses to aggress against others.
The deterministic belief essentially says that the person
could not act otherwise, which resembles a standard form
of excuse (I couldnt help it) and thus might encourage
people to act in short-sighted, impulsive, selfish ways.
To emphasize the inner shift in state, two of the present experiments used a manipulation other than the
Crick (1994) essay because the essay itself might in principle have a variety of possible effects. Following the
newer procedure developed by Vohs and Schooler
(2008), we used a Velten-style procedure that involved
having participants read and internalize a series of statements that explicitly promoted or rejected the sense of
personal choice and responsibility for ones actions. This
procedure has the additional merit of contrasting the
promotion of belief in free will in one condition with
belief in determinism in a second condition (and, in
Experiment 1, a neutral control). These inductions combined statements about the link between freedom and
personal choice and about ostensible scientific findings
regarding the causation of human behavior by genetic,
environmental, and brain processes. Experiment 2
measured disbelief in free will instead of manipulating
it. We predicted that disbelief in free will, whether
manipulated or measured, would decrease helping and
increase aggression.

EXPERIMENT 1
Experiment 1 provided a simple and preliminary
test of the hypothesis that disbelief in free will reduces

Downloaded from psp.sagepub.com at University of Tokyo on September 3, 2011

262

PERSONALITY AND SOCIAL PSYCHOLOGY BULLETIN

prosocial tendencies. Participants completed a Veltenstyle procedure in which (by random assignment) one
third of them internalized statements related to disbelief
in free will, another one third internalized states related
to a belief in free will, and the remainder pondered neutral statements. Participants then reported how likely
they would be to offer help to people depicted in several
scenarios. We predicted that disbelief in free will would
lead to lower helping intentions compared to participants
in the condition that encouraged belief in free will.
The neutral condition was of more interest than many
standard control conditions. A simple prediction would
be that its effects would be intermediate between the two
manipulated belief conditions. However, if the neutral
control differed from only one of them, that pattern
would suggest that the other belief was fairly equivalent
to what people normally believe. Our general assumption was that belief in free will promotes socially desirable and harmonious behavior, and so people generally
are socialized to believe in free will. On that basis, we
expected that the induced free will condition and the
neutral control conditions would yield similar results. In
contrast, if people mostly disbelieve in free will, the control condition should differ only from the profree will
condition.
Method
Participants. Participants were 70 undergraduate
students (30 female) who participated to fulfill a course
requirement. Six participants expressed suspicion about
the purpose of the study. Their data were excluded,
leaving a total of 64 participants.
Procedure. Participants were run together in a
large university classroom. They were asked to sit so
that no two people were adjacent to one another.
After giving informed consent, participants were
given written and oral instructions for the sentencereading procedure.
The free will belief manipulation was adapted from
Vohs and Schooler (2008). Participants were given a
packet containing 15 pages, with one sentence on each
page. Participants were randomly assigned to read sentences in support of either free will or determinism or, in
the neutral control condition, sentences that had no relevance to free will or determinism. After receiving their
packets, participants were given written and recorded
audio instructions explaining that they would be
required to read the packet 1 page at a time. Participants
were told that every time they heard a tone, they should
move on to read the next page. Tones occurred once
every minute, so that it took a total of 15 min to read the
full set of sentences.

The determinism sentences included Science has

demonstrated that free will is an illusion; Like everything else in the universe, all human actions follow from
prior events and ultimately can be understood in terms of
the movement of molecules; and All behavior is determined by brain activity, which in turn is determined by a
combination of environmental and genetic factors. The
free will sentences included I demonstrate my free will
every day when I make decisions, I am able to override
the genetic and environmental factors that sometimes
influence my behavior, and I have feelings of regret when
I make bad decisions because I know that ultimately I am
responsible for my actions. Neutral statements emphasized geography, scientific facts, and world events, such as
Oceans cover 71% of the earths surface, Alkaline
power cells generally work longer than ordinary batteries,
and Pocket calculators became common items only after
1970.
Participants were randomly assigned to one of the
three conditions (free will, determinism, or neutral control). After reading the sentences, participants completed the Brief Mood Introspection Scale (BMIS;
Mayer & Gaschke, 1988).
Participants then read six hypothetical scenarios in
which they had the opportunity to help others (e.g., giving money to a homeless person, allowing a fellow
classmate to use ones cellular phone). Participants indicated the likelihood that they would help in that situation (based on how they would behave in each situation
at the present moment) using a scale from 1 (not at all
likely) to 9 (very likely). Responses to the six scenarios
were averaged to form the dependent measure of helping. All participants were probed for suspicion using a
funneled debriefing procedure (see Bargh & Chartrand,
2000) and given course credit.
Results
Likelihood of helping. ANOVA on the combined helpfulness index revealed significant variation among conditions, F(2, 61) = 3.23, p < .05. Pairwise comparisons
indicated that participants in the determinism condition
(M = 5.33, SD = 1.52) were less willing to help than were
participants in the free will condition (M = 6.27, SD =
1.19), F(1, 61) = 4.84, p = .03, and less helpful than participants in the neutral control condition (M = 6.23, SD =
1.28), F(1, 61) = 4.99, p < 0.03. There was no difference
between the neutral control and the free will conditions,
F < 1, ns.
Mood and arousal. The effect of the manipulation did
not appear to be attributable to mood or arousal (as
assessed by the BMIS). ANOVA revealed no significant
variation among conditions on mood valence, F(2, 61) =
1.91, ns, or arousal, F(2, 61) = 1.94, ns.

Downloaded from psp.sagepub.com at University of Tokyo on September 3, 2011

Baumeister et al. / PROSOCIAL BENEFITS OF FEELING FREE

Validation Studies
To verify the manipulations effectiveness and check
on possible demand characteristics, we performed two
separate validation studies. Participants in the first were
54 students in a psychology class. They underwent the
free will feelings induction procedure (used in
Experiments 1 and 3) and then rated several reactions
on scales ranging from 1 (agree) to 7 (disagree). On the
item, I feel I am free to choose whatever I want to do
in my life, ANOVA revealed significant variation
among conditions, F(2, 51) = 3.87, p < .05, and planned
comparisons confirmed that participants who had read
the free will induction agreed with that more strongly
than those who had read the determinism induction,
F(1, 51) = 6.36, p < .02. Those who read the neutral
control statements agreed with that statement more
strongly than those who had read the determinism
induction condition, F(1, 51) = 4.22, p < .05, but neutral control participants did not differ from participants
in the free will induction condition, F < 1, ns. Thus, the
manipulation did induce some differences in feelings of
free will, and in particular it appears that the differences
were due to reduction of such feelings by the determinism condition, whereas the free will condition appears
to be similar to the way people normally feel.
Further results contradicted a demand characteristic
explanation for the results of Experiments 1 (decreased
helping) and 3 (increased aggression). There were no differences between conditions on the item The person
who wrote those statements probably would want me to
be kind and helpful, F(2, 51) = 1.32, ns. Likewise, there
were no differences on The person who wrote those
statements would probably want me to be mean and
cruel, F < 1, ns. (This item was chosen to anticipate
Experiment 3 and its aggression measure.) Apparently,
the manipulations were not perceived as directly fostering
expectations about either helpfulness or aggressiveness.
The second study was inspired by journal reviews of
an earlier version of this manuscript. It sought to examine the effect of the manipulation on free will beliefs,
perceived accountability, and feelings of agency.
Participants were 53 students enrolled in an introductory psychology course who came individually into the
lab. Participants underwent the same free will belief
manipulation as in Experiment 3 that included free will
and determinism conditions (from Experiment 1) but no
neutral control condition. Participants then completed
the same version of the Free Will and Determinism scale
(FAD; Paulhus & Margesson, 1994) used in Study 2
and gave ratings from 1 (agree) to 7 (disagree) for several items related to feelings of accountability (e.g., I
am held accountable for my actions) and agency (e.g.,
Right now, I feel active; descriptors were taken from

263

Campbell, Bosson, Goheen, Lakey, & Kernis, 2007).

The agency items together had an alpha of .87, indicating that they hang together well. For the accountability
items, the alpha was somewhat lower (.55) though
acceptable, especially as a dependent variable. We
report analyses both for the combined scales and for
individual items.
FAD scores were coded such that higher numbers
indicated greater belief in free will. ANOVA results
showed that the free will condition exhibited higher
FAD scores (M = 55.22, SD = 5.75) than the determinism condition (M = 52.30, SD = 3.20), F(1, 52) = 5.15,
p < .03. Thus, the manipulation induced the predicted
difference in free will beliefs such that those in the determinism condition showed lower endorsement of free
will beliefs than those in the profree will condition.
Further analyses showed no effect of the manipulation on perceptions of accountability or feelings of
agency. There were no differences between the conditions on the accountability items I feel like I have to
explain why I do things, I am held accountable for
my actions, or I dont need to explain the reasons for
my actions to anyone; all Fs < 1, ns. There were also
no differences between conditions in reports of feeling
assertive, active, energetic, outspoken, dominant, or
enthusiastic, all Fs < 2.31, ps > .135. Composite scores
of the accountability items and of the agency items also
yielded no differences between the two conditions, Fs <
1, ns. These data suggest that the manipulation affected
participants beliefs in free will but did not directly
affect feelings of accountability or agency. They also
speak against any idea that the disbelief manipulation
induced feelings of learned helplessness.
Discussion
Prosocial tendencies (as indicated by willingness to
help) were reduced among participants who were
induced to believe in determinism and disbelieve in free
will. They were less willing to help across an assortment
of situations and opportunities, as compared to participants who were induced to believe in free will and as
compared to a neutral control group.
We outlined three hypotheses about why the manipulation would have these effects on behavior and considered several alternative explanations. The effects
appeared to be due to altered beliefs in free will versus
deterministic inevitability. Mood reports ruled out mediation by emotional state, including both mood valence
and arousal. The first validation study indicated that the
manipulation was effective in altering belief in free will
and determinism and did not convey demand characteristics about expected responses. The second validation
study indicated that the manipulation did not alter

Downloaded from psp.sagepub.com at University of Tokyo on September 3, 2011

264

PERSONALITY AND SOCIAL PSYCHOLOGY BULLETIN

feelings of accountability and personal responsibility, nor

did it bring about changes in conscious feelings of being
active and wanting to exert agentic volition. Taken
together, these results point toward the conclusion that
disbelief in free will serves as a subtle prime that encourages people to respond automatically (here, by selfishly
or lazily spurning the opportunity to help) rather than
exerting self-control (possibly because exerting selfcontrol is strenuous and depletes energy resources).
Promoting belief in free will had no effect in comparison to the (neutral) control condition. This suggests
that promoting belief in free will does not change
people away from their normal state and attitudes
unlike promoting disbelief in free will, which did cause
a change. The implication is that most ordinary people
believe in free will under normal circumstances. This
pattern likewise speaks against the hypothesis that the
free will induction specifically gives rise to a personal
feeling of active energy or a desire to exert control.
EXPERIMENT 2
Experiment 2 instituted two important changes.
First, it shifted to measurement of individual differences. Whereas Experiment 1 manipulated belief in
determinism and free will, Experiment 2 measured how
much participants chronically disbelieved free will. This
was meant to provide converging evidence that manipulated and measured disbelief in free will would have
similar effects on prosocial responses.
Second, Experiment 2 measured behavioral commitment to provide actual help rather than relying on selfreported willingness to help in hypothetical scenarios. It
used the well-validated Katie Banks helping paradigm (Batson et al., 1997; Maner et al., 2002) in which
participants can volunteer to help a college student
whose parents were killed recently. Although some
work has shown convergence between responses to scenarios and actual responses (Robinson & Clore, 2002),
other work has shown some discrepancies between
imagined helping responses and actual helping behavior
(West & Brown, 1975). Because the present procedure
involved making a commitment to provide actual help,
it is considerably closer to actual behavior than mere
hypothetical responses to imaginary scenarios. Based on
the findings of Experiment 1, we predicted that higher
levels of dispositional belief in free will would be lead to
greater helpfulness.
Method
Participants. Participants were 52 undergraduates
(31 women, 20 men, 1 not reported) who participated
in exchange for partial course credit.

Materials and procedure. Participants arrived at the

laboratory individually. They were told simply that the
study was exploring the relationship between language
comprehension and personality. Participants completed a
short version of the FAD (Paulhus & Margesson, 1994)
and the Positive and Negative Affect Schedule (PANAS;
Watson, Clark, & Tellegen, 1988). The PANAS is a widely
used, standard measure of emotional state.
The FAD is scale developed to assess individual differences in belief in free will and determinism. It has
four dimensions, or subscales, as follows, with sample
items. First, belief in fate reflects the inability of individuals to control their destiny (e.g., What will be, will
betheres not much you can do about it and I
believe that my future has already been pre-determined
by fate). Second, belief in scientific causation emphasizes both biological (e.g., Your genes determine your
future and Peoples biological makeup influences
their talents and personality) and environmental (e.g.,
Bad behavior is caused by bad life circumstances)
causes of human behavior. Third, belief in randomness
depicts human life as dependent on chance events (e.g.,
Chance events seem to be the major cause of human
history and No one can predict what will happen in
this world). Last, belief in free will specifically refers to
peoples ability to decide and control their own behavior and outcomes, with emphasis on personal responsibility for ones actions (e.g., People have complete
control over the decisions they make and Criminals
are totally responsible for the bad things they do).
Responses to each item range from 1 to 5. Paulhus and
Margesson (1994, 2008) report alphas for the most
recent version of the subscales ranging from .84 to .61,
and 2-month retest reliabilities ranging from .79 to .90.
In the 16-item short version used in the present study,
we found alpha of .55 (and .66 in other investigations
with the same scale), which is close to the alpha of .65
for the full scale according to Paulhus and Margesson
(2008). Retest reliability after delays varying at random
from 6 to 10 weeks were found to be .76, indicating
that these beliefs are fairly stable.
After completing the FAD and PANAS, the experimenter informed participants that they would listen to
one of six possible radio broadcasts. Participants rolled
a die, ostensibly to determine to which of the broadcasts
they would listen to. This procedure was rigged so that
all participants listened to the same radio interview that
supposedly aired on the radio. During the interview, a
woman named Katie Banks described how her parents
had been killed in a car accident and how she was now
solely responsible for the care of her siblings. Because of
that responsibility, she said she would have to drop out
of college and take a job unless she could find someone
who could help her financially or with the care of her

Downloaded from psp.sagepub.com at University of Tokyo on September 3, 2011

Baumeister et al. / PROSOCIAL BENEFITS OF FEELING FREE

siblings. After listening to the interview, participants
completed measures tapping their cognitive and affective responses to the interview.
The experimenter then informed participants that the
study was over. They received a bogus debriefing and
received a credit slip for participating. Before leaving,
the experimenter explained that the faculty supervisor
for the study (Dr. Edmunds) had requested that all participants who heard the Katie Banks interview be given
the opportunity to volunteer to help her. Participants
were given letters from Dr. Edmunds and Katie that
described how participants could volunteer their time
helping Katie by performing a variety of tasks (e.g.,
stuffing envelopes). The experimenter gave participants
a volunteer form on which participants indicated how
many hours (0 to 9 or more) they were willing to help.
After indicating how many hours they would help, participants were given a funneled debriefing to check for
suspicion and were dismissed.
Results and Discussion
We predicted that chronic disbelief in free will would
predict unhelpful behavior toward Katie Banks. Because
the majority of participants (71%) did not volunteer
any help, we analyzed the helping data as both a continuous measure (0-9 or more hours) and a dichotomous measure (0 = did not volunteer to help, 1 =
volunteered to help). FAD responses were coded so that
higher scores indicated greater disbelief in free will.
Belief in free will was positively associated with helping behavior. Results showed that disbelief in free will
predicted a lower number of hours for which participants volunteered, = .30, t = 2.24, p < .03. Treating
helping as a dichotomous measure yielded similar
results. Disbelief in free will was associated with a lower
tendency to volunteer any help at all, B = .15, SE = .07,
Wald = 4.95, p < .03. These findings suggest that
chronic disbelief in free will relates to a lower likelihood
of helping another person. These results thus converge
with what Experiment 1 found.
Because emotion can play a role in helping behavior
(e.g., Cialdini & Kenrick, 1976; Maner et al., 2002), we
conducted further analyses to test possible emotion
effects. Neither positive affect nor negative affect was
associated with helping behavior, both ps > .26. When
we controlled for positive affect and negative affect, the
relationship between chronic disbelief in free will and
helping remained significant when helping was treated
as both a continuous measure, = .37, t = 2.66, p =
.01, and a dichotomous measure, B = .20, SE = .08,
Wald = 6.41, p = .01. Thus, the link between chronic
disbelief in free will and reduced helping behavior was
not due to positive or negative affect.

265

EXPERIMENT 3
Experiment 3 tested the hypothesis that disbelief in
free will would influence aggression toward a seemingly
innocent target. Most aggression research measures
retaliation for an insult, frustration, or other angry
provocation, and aggressive behavior under those circumstances can be affected by perceived inappropriateness of the others behavior. Aggression toward an
innocent target is, however, generally perceived as totally
inappropriate. To increase the likelihood of getting any
aggression toward an innocent target, we sought (unsuccessfully, it turned out) to stimulate feelings of social
exclusion and rejection among some participants, based
on previous findings that rejected people become slightly
more aggressive toward innocent bystanders (Twenge,
Baumeister, Tice, & Stucke, 2001; see also Leary,
Kowalski, Smith, & Phillips, 2003).
Experiments 1 and 2 found that state and trait beliefs
in determinism were related to reduced helping behavior
or tendencies, but those results could be taken to show
that disbelief in free will makes people passive and lazy
rather than indicating a reduction in prosocial behavior.
The aggression measure in the current experiment provided an opportunity to rule out the influence of this
confound. Aggression is an automatic and antisocial
response (DeWall et al., 2007) that, if anything, requires
more action than being nonaggressive. The idea that the
free will manipulation makes people passive would lead
to the prediction that it would make people less aggressive. In contrast, our hypothesis was that exerting control takes energy and that disbelief in free will makes
people reluctant to expend their energy in acts of selfcontrol. On that basis, we predicted that disbelief in free
will would produce an increase in aggression.
The aggression measure was adapted from
Lieberman, Solomon, Greenberg, and McGregor (1999)
and consisted of giving a hot and spicy food stimulus to
another participant who had expressed a severe dislike
for such foods. Aggression is defined as providing aversive stimulation to someone who does not want it and
is motivated to avoid it (Baron & Richardson, 1994),
and the hot sauce procedure was explicitly designed to
satisfy those requirements. The main prediction was
that aggression would be highest among participants
who had been induced to think a series of thoughts that
emphasized a lack of free will.
Method
Participants. Participants were 56 undergraduates
(45 female) who participated to fulfill a course requirement. Seven participants expressed suspicion about the
social exclusion manipulation and/or the presence of

Downloaded from psp.sagepub.com at University of Tokyo on September 3, 2011

266

PERSONALITY AND SOCIAL PSYCHOLOGY BULLETIN

another person during the aggression measure. Their

data were excluded, leaving a total of 49 participants in
the final analysis.
Procedure. Participants arrived at the laboratory in
single-sex groups of 4 to 6 people. After signing consent
forms, participants received name tags on which they
wrote their first names. They were then given both written and oral instructions to talk for 10 min using a set
of printed questions as a guide (the questions were
taken from the closeness induction task developed by
Sedikides, Campbell, Reeder, & Elliot, 1999). After 10
min, the experimenter led the participants to individual
rooms where they completed a demographic form and
an additional page with the following instructions: We
are interested in forming groups in which the members
like and respect each other. Below, please name the two
people (out of those you met today) you would most
like to work with. The experimenter collected these
sheets and told the participants he would return with
their group assignments.
Participants were assigned randomly to be either
accepted or rejected by the other participants. The
accepted participants were told, I have good news for
youeveryone chose to work with you. The rejected
participants were told, Unfortunately, no one chose to
work with you. This procedure was adapted from
Leary, Tambor, Terdal, and Downs (1995) and Nezlek,
Kowalski, Leary, Blevins, and Holgate (1997). The
experimenter then explained to all participants that the
groups could not be worked out as usual, so they would
be taking part in a taste preferences study with a new
partner. It was explained that the new person was
making up an experiment that required 2 participants.
This person was always referred to as being the same
sex as the participant. Participants then completed the
BMIS mood measure and a taste preferences scale that
assessed liking for foods of six types (salty, spicy, dry,
sweet, sour, and creamy) on a scale from 1 (extreme disliking) to 21 (extreme liking).
The free will belief manipulation was the same as
used in Experiment 1 (from Vohs & Schooler, 2008)
except that we did not include a neutral belief control
condition. Participants read through 15 sentences, 1 per
minute, either asserting human freedom of action or
deterministic inevitability.
The aggression measure was adapted from work by
Lieberman et al. (1999). After completing the reading
task, participants were asked to prepare some food
samples for their partner. Participants were given a taste
preferences form, which they were told had been filled
out by their partner. In reality, all participants received
an identical taste preferences form, which had been
filled out to highlight the partners strong dislike for

spicy food. It rated all other food types in the moderate

range (between 7 and 15 out of 21) but rated spicy
foods at 3. For the first food sample, participants were
given three crackers and a can of processed cheese. They
were told to add cheese to the crackers as they saw fit.
After creating this initial food sample, participants were
given three tortilla chips and a jar of salsa. The salsa
was ostensibly spicy, and the jar had the words hot
printed on it in several places. Participants were told to
add as much salsa as they wanted. Throughout the food
preparation procedure, it was stressed that the partner
would be required to eat all of the food on the plate.
Food samples were measured both before and after they
were given to the participants, and the difference provided the dependent measure of aggression (milligrams
of salsa used).
After completing the study, participants were given a
funneled debriefing to check for suspicion. It was also
explained that the acceptance versus rejection feedback
had been randomly assigned and that none of the participants actually worked together. Rejected participants were assured that some people had in fact chosen
to work with them and that the feedback was not true.
All participants were thanked, given course credit, and
dismissed.
Results and Discussion
The main measure of aggression was the quantity of
hot sauce given by the participant to the other person
who had explicitly expressed a dislike for such tastes.
Gender was included as a covariate in the analyses
because there was a main effect such that males were
more aggressive than females, F(1, 48) = 7.38, p < .01,
although gender did not interact with the other variables, F < 1, ns.
ANCOVA revealed that participants who had read
the deterministic sentences gave their partners more of
the unwelcome hot sauce (M = 17.8 mg, SD = 16.3)
compared to participants who read the sentences supporting free will (M = 9.4, SD = 11.6), F(1, 48) = 6.95,
p = .01. Thus, inducing disbelief in free will led to more
aggression as compared to inducing belief in free will.
The findings confirmed that encouraging a deterministic disbelief in free will resulted in higher aggression
than encouraging belief in free will.
One ambiguity with the hot sauce measure is that it
could be interpreted as wanting to give the other person
more food in total, in which case it could express a
prosocial rather than an aggressive intent. If believing in
determinism made people want to give the partner more
food overall, this would presumably be reflected in giving the partner more cheese as well as more salsa.1
Hence, we analyzed the amount of cheese allocated.

Downloaded from psp.sagepub.com at University of Tokyo on September 3, 2011

Baumeister et al. / PROSOCIAL BENEFITS OF FEELING FREE

ANCOVA found no significant difference in cheese portions between the participants who had undergone the
determinism induction, F(1, 48) = 1.87, ns. Thus,
apparently the impact of the belief manipulation was
specific to the aggressive (salsa) response.
The acceptance versus rejection manipulation did not
contribute significantly to aggressive behavior, either as
main effect, F < 1, ns, or in interaction with the free will
manipulation, F(1, 45) = 1.19, ns. The failure to replicate previous work deserves comment. The delay
between the rejection manipulation and the aggression
measure may well have been a factor. Past laboratory
studies finding a link between rejection and aggression
have generally administered the aggression measure
immediately after the rejection manipulation, whereas
in the present study there was a delay of more than half
an hour, indeed one filled with other impactful tasks.
The importance of temporal proximity between rejection and aggression has been emphasized by Arnold,
Homrok, Ortiz, and Stowe (1999). Blackhart, Eckel,
and Tice (in press) have likewise shown that the subjective and physiological effects of this laboratory manipulation dissipate during the first half hour. It may also
be that the hot sauce measure of aggression used here is
less sensitive than the noise blast procedure used by
Twenge et al. (2001; likewise the exclusively male sample of school shooters discussed by Leary et al., 2003)
and therefore failed to register increased aggressiveness.
Also, possibly the higher proportion of female participants in this study than in that previous study may have
reduced the ability to detect small effects, insofar as
females are less aggressive than males.
We measured mood with the BMIS after the rejection
manipulation (but before the free will manipulation). As in
much previous work, the rejection manipulation had no effect
on either mood valence or mood arousal, both Fs < 1, ns.

GENERAL DISCUSSION
The present results support the view that belief
in free will is a valuable support for prosocial behavior,
as suggested by Vohs and Schooler (2008). Those
researchers showed that disbelief in free will made
people more willing to cheat on a test. Our findings
extend this pattern into two major categories of prosocial and antisocial behavior. Specifically, we found that
inducing people to disbelieve in free will led to an
increase in aggression and a reduction in willingness to
help. Individuals who were chronically high in disbelief
in free will were also less likely to help another person
in distress as compared to people who were dispositionally skeptical or rejecting of free will. All these results
were found to be independent of mood and emotion.

267

Experiment 1 found that willingness to help was the

same in a neutral control group as among participants
who were induced to believe in free will. This presumably indicates that most participants generally have
such a belief in free will already. Apparently, inducing
disbelief is more a departure from the norm than inducing belief in free will.
The broader implication is that many people in
Western culture share a belief in human freedom of
action and that, moreover, human society benefits from
such a belief. (Indeed, we suspect that most cultures will
have found beliefs in free will to be socially beneficial
and hence will tend to favor and promote those beliefs.)
Volition and self-control require the person to expend
energy, and these expenditures enable them to act
prosocially. Apparently disbelief in free will subtly
reduces peoples willingness to expend that energy.
Hence, disbelief in free will serves as a cue to act on
impulse, a style of response that promotes selfish and
impulsive actions such as aggressing and refusing to
help. Some philosophical analyses may conclude that a
fatalistic determinism is compatible with highly ethical
behavior, but the present results suggest that many
laypersons do not yet appreciate that possibility.
Of course, the present results have nothing to say
about whether free will is an objective reality. All of our
findings could be perfectly valid even if free will is a
complete illusion. It would hardly be the only one. Our
results do indicate, however, that belief in free will contributes to socially beneficial actions and outcomes.
Indeed, if it is an illusion, those benefits may well
explain why society continues to find it useful to promote it.

NOTE
1. To be sure, giving more cheese would not be unambiguously
prosocial: One might provide more cheese simply to provide a bigger
vehicle for forcing more salsa on the hapless partner.

REFERENCES
Arnold, D. H., Homrok, S., Ortiz, C., & Stowe, R. M. (1999).
Direct observation of peer rejection acts and their temporal relation with aggressive acts. Early Childhood Research Quarterly,
14, 183-196.
Bargh, J. A., & Chartrand, T. L. (2000). The mind in the middle: A
practical guide to priming and automaticity research. In H. T. Reis
& C. M. Judd (Eds.), Handbook of research methods in social and
personality psychology (pp. 253-285). New York: Cambridge
University Press.
Baron, R. A., & Richardson, D. R. (1994). Human aggression (2nd
ed.). New York: Plenum.
Baumeister, R. F., Bratslavsky, E., Muraven, M., & Tice, D. M. (1998).
Ego depletion: Is the active self a limited resource? Journal of
Personality and Social Psychology, 74, 1252-1265.

Downloaded from psp.sagepub.com at University of Tokyo on September 3, 2011

268

PERSONALITY AND SOCIAL PSYCHOLOGY BULLETIN

Batson, C. D., Sager, K., Garst, E., Kang, M., Rubchinsky, K., &
Dawson, K. (1997). Is empathy-induced helping due to self-other
merging? Journal of Personality and Social Psychology, 73, 495-509.
Blackhart, G. C., Eckel, L. A., & Tice, D. M. (2007). I know they
really like me: Defensiveness as a moderator of neuroendocrine
responses to social rejection by peers. Biological Psychology, 75,
267-276.
Campbell, W. K., Bosson, J. K., Goheen, T. W., Lakey, C. E., &
Kernis, M. H. (2007). Do narcissists dislike themselves deep
down inside? Psychological Science, 8, 227-229.
Cialdini, R. B., & Kenrick, D. T. (1976). Altruism as hedonism: A
social development perspective on the relationship of negative
mood and helping. Journal of Personality and Social Psychology,
34, 907-914.
Crick, F. (1994). The astonishing hypothesis: The scientific search for
the soul. New York: Simon & Schuster.
DeWall, C. N., Baumeister, R. F., Gailliot, M. T., & Maner, J. K. (in
press). Depletion makes the heart grow less helpful: Helping as a
function of self-regulatory energy and genetic relatedness.
Personality and Social Psychology Bulletin.
DeWall, C. N., Baumeister, R. F., Stillman, T. F., & Gailliot, M. T.
(2007). Violence restrained: Effects of self-regulatory capacity and
its depletion on aggressive behavior. Journal of Experimental
Social Psychology, 43, 62-76.
Gailliot, M. T., Baumeister, R. F., DeWall, C. N., Maner, J. K., Plant,
E. A., Tice, D. M., et al. (2007). Self-control relies on glucose as a
limited energy source: Willpower is more than a metaphor. Journal
of Personality and Social Psychology, 92, 325-336.
Kane, R. (2002). Oxford handbook of free will. New York: Oxford
University Press.
Kant, I. (1967). Kritik der reinen Vernunft [Critique of pure reason]. Hamburg,
Germany: Felix Meiner Verlag. (Original work published 1787)
Leary, M. R., Kowalski, R. M., Smith, L., & Phillips, S. (2003).
Teasing, rejection, and violence: Case studies of the school shootings. Aggressive Behavior, 29, 202-214.
Leary, M. R., Tambor, E. S., Terdal, S. K., & Downs, D. L. (1995).
Self-esteem as an interpersonal monitor: The sociometer hypothesis. Journal of Personality and Social Psychology, 68, 518-530.
Lieberman, J. D., Solomon, S., Greenberg, J., & McGregor, H. A.
(1999). A hot new way to measure aggression: Hot sauce allocation. Aggressive Behavior, 25, 331-348.
Maner, J. K., Luce, C. L., Neuberg, S. L., Cialdini, R. B., Brown, S.,
& Sagarin, B. J. (2002). The effects of perspective taking on
helping: Still no evidence for altruism. Personality and Social
Psychology Bulletin, 28, 1601-1610.

Mayer, J. D., & Gaschke, Y. N. (1988). The experience and meta-experience

of mood. Journal of Personality and Social Psychology, 55, 102-111.
Nezlek, J. B., Kowalski, R. M., Leary, M. R., Blevins, T., & Holgate, S.
(1997). Personality moderators of reactions to interpersonal rejection: Depression and trait self-esteem. Personality and Social
Psychology Bulletin, 23, 1235-1244.
Paulhus, D. L., & Margesson, A. (1994). Free Will and Determinism
(FAD) scale. Unpublished manuscript, University of British
Columbia, Vancouver, Canada.
Paulhus, D. L., & Margesson, A. (2008). Revised Free Will and
Determinism (FAD) scale. Unpublished manuscript, University of
British Columbia, Vancouver, Canada.
Robinson, M. D., & Clore, G. L. (2002). Episodic and semantic
knowledge in emotional self-report: Evidence for two judgment
processes. Journal of Personality and Social Psychology, 83,
198-215.
Sedikides, C., Campbell, W. K., Reeder, G. D., & Elliot, A. J. (1999).
The relationship closeness induction task. Representative Research
in Social Psychology, 23, 1-4.
Twenge, J. M., Baumeister, R. F., Tice, D. M., & Stucke, T. S. (2001).
If you cant join them, beat them: Effects of social exclusion on
aggressive behavior. Journal of Personality and Social Psychology,
81, 1058-1069.
Vohs, K. D., Baumeister, R. F., Schmeichel, B. J., Twenge, J. M.,
Nelson, N. M., & Tice, D. M. (2008). Making choices impairs
subsequent self-control: A limited resource account of decision
making, self-regulation, and active initiative. Journal of
Personality and Social Psychology, 94, 883-898.
Vohs, K. D., & Schooler, J. (2008). The value of believing in free will:
Encouraging a belief in determinism increases cheating.
Psychological Science, 19, 49-54.
Watson, D., Clark, L. A., & Tellegen, A. (1988). Development and
validation of brief measures of positive and negative affect: The
PANAS scales. Journal of Personality and Social Psychology, 47,
1063-1070.
Wegner, D. M. (2002). The illusion of conscious will. Cambridge,
MA: MIT Press.
West, S. G., & Brown, T. J. (1975). Physical attractiveness, the severity
of emergency and helping: A field experiment and interpersonal simulation. Journal of Experimental Social Psychology,
11, 531-538.
Received February 18, 2008
Revision accepted August 26, 2008

Downloaded from psp.sagepub.com at University of Tokyo on September 3, 2011